1. Field of the Invention
The present invention relates to a motion detecting device, a motion detecting method, an imaging device, and a monitoring system. In particular, it relates to a motion detecting device for detecting a motion of a subject from a video signal, a motion detecting method therefor, an imaging device including the motion detecting device, and a monitoring system using the imaging device.
2. Description of Related Art
A motion detecting device, specifically, a motion detecting device which detects a motion of a subject from a video signal, is used for monitoring a suspicious person in a monitoring system and as a trigger of a DVR (Digital Video Recording). For example, the monitoring system intends to alleviate a load of a monitoring person and to effectively record (record a necessary scene only), and controls to analyze a video outputted from a monitoring camera (in this case, detect motion within an image area) to issue a warning to the monitoring person or to record a recoding device.
As a motion detecting method of the motion detecting device, an interframe difference method and a background difference system are generally known. The interframe difference method is a technique in which an interframe brightness difference between a video signal outputted from an image sensor and a video signal delayed by one frame is measured, and the brightness difference is compared with a previously set detection threshold value, thereby detecting a presence or absence of motion. The background difference method is a technique in which a background image is produced by an arithmetic operation from a video signal outputted from an image sensor, a brightness difference between the inputted video signal and the background image (background model) is measured, and the brightness difference is compared with a previously set detection threshold value, thereby detecting a presence or absence of motion.
The motion detecting device using these detecting methods generally divides an image area into N×M of blocks, evaluates an average value thereof, and decreases a resolution, thereby alleviating an amount of arithmetic operation. The background image may be created by the following expression (1):B(m,n,t)=(1−α)B(m,n,t−1)+αI(m,n,t)  (1),where α denotes a time constant of 0≦α≦1, B (m,n,t) denotes a background image of coordinates (m, n) at a time t, and I(m,n,t) denotes an input image of coordinates (m, n) at the time t.
In a usual camera system, a brightness signal is integrated to find out a brightness level within the image area; an aperture (F-number), a shutter speed, and an AGC (Automatic Gain Control) gain or the like are controlled such that the integrated brightness value approaches a control target value; and AE (Auto Exposure) control in which an exposure is automatically determined is performed.
In a monitoring system using the camera system in which such AE control is performed is used as the monitoring camera, when the arithmetic operation of the expression (1) is performed in detecting a motion, a brightness change resulting from an actual motion of a subject and a brightness change resulting from the AE control cannot be separated (or estimated). As a result, the brightness change resulting from the AE control also is detected as the motion of the subject, so that a false alarm is issued. Such a false alarm increases labor of the monitoring person or causes a recording even if no motion has occurred, which leads to prevention of an effective recording.
To separate the brightness change resulting from the actual motion of the subject and the brightness change resulting from the AE control, it may be considered to adopt a method in which a threshold value of the motion detection is set large in advance. However, when the measure in which the detected threshold value is set large in advance is adopted, there is a demerit in that sensitivity of the detection decreases.
Therefore, in the existing technologies, based on a concept that edges exist in the actual motion, an edge detecting circuit for detecting edge portions of the subject from a video signal outputted from the monitoring camera and a video signal delayed by one frame is used to increase detection precision of the motion detection based on a detection result of the edge detecting circuit (for example, see Japanese Unexamined Patent Application Publication No. 2001-8228, hereinafter referred to as “Patent Document 1”).